Quant Market Brief — September 29, 2026: AI Oversight & Options Quote Controls

Robert Henson

— 6 min read
Share

Research cutoff: September 29, 2026, 6:34 a.m. CT (11:34 UTC). Premarket readings are indicative or non-consolidated where noted; other observations are prior-close or latest verified publications.

1. White House convenes AI leaders on oversight

Abstract frontier-AI governance, oversight, and evaluation illustration

Credit: original Market State Lab category illustration; not an event photograph.

Fact: President Trump is scheduled to meet AI industry leaders Tuesday, Reuters reported September 29. Speaker Mike Johnson described innovation and oversight as priorities. Expected participants include Meta’s Mark Zuckerberg and Anthropic’s Dario Amodei. Reuters, September 29

Inference: Governance expectations may change independently of model capability.

Uncertainty: The meeting had not occurred at the cutoff; attendance, proposals, and outcomes were not settled.

Why it matters: Maintain permission inventories, reproducible evaluations, independent shutdown controls, and human approval for irreversible MATLAB-agent actions.

2. MIAX changes option quote-width settings for Q4

Abstract options quote validation, bid-ask spreads, and exchange controls illustration

Credit: original Market State Lab category illustration; not an event photograph.

Fact: MIAX’s September 28 notice changes maximum valid bid/ask differentials for certain symbols on MIAX Options and MIAX Emerald from October 1 through December 31, unless withdrawn earlier. MIAX, September 28

Inference: Exchange-valid quotes and economically acceptable execution are separate tests.

Uncertainty: Affected-symbol thresholds were not verified here; do not assume SPY or QQQ is affected. Actual spreads and fills remain unknown.

Why it matters: Map circular settings by symbol, venue, and effective date. Replay quote validation and monitor rejections, cancellations, and fills; permitted width is not an acceptable cost budget.

3. Futures steady as technology rebounds against oil and yield pressure

Abstract SPY, QQQ, Treasury yields, and opening-auction risk illustration

Credit: original Market State Lab category illustration; not an event photograph.

Fact: At 4:06 a.m. CT, Dow futures were down 0.07%, S&P 500 futures 0.04%, and Nasdaq-100 futures up 0.06%. Reuters described a technology rebound constrained by oil and elevated yields. Reuters, September 29

Inference: A small technology bounce does not establish a durable reversal.

Uncertainty: Futures were indicative; consolidated spreads, depth, auction imbalances, and dealer gamma were unavailable.

Why it matters: Separate premarket, auction, and first-30-minute books. Gate participation on fresh NQ–ES strength, yields, breadth, and executable quote quality.

4. Cook warns AI investment can broaden inflation before productivity helps

Abstract AI investment, labor demand, energy, and inflation transmission illustration

Credit: original Market State Lab category illustration; not an event photograph.

Fact: In a September 28 speech, Fed Governor Lisa Cook said AI investment can increase demand for construction labor and energy, spreading price pressure beyond the sector. Productivity gains may modestly restrain inflation over coming years without offsetting nearer-term pressure. Federal Reserve, September 28

Inference: AI capital spending can initially act as a demand shock before supply benefits arrive.

Uncertainty: This is one policymaker’s assessment, not an FOMC commitment; the scale and timing of productivity gains are uncertain.

Why it matters: Model AI investment demand, productivity, energy pass-through, and labor mismatch separately. Keep future policy paths conditional on incoming data.

5. AstraZeneca invests in Summit cancer-development collaboration

Abstract biotechnology research, strategic investment, and clinical-development collaboration illustration

Credit: original Market State Lab category illustration; not an event photograph.

Fact: AstraZeneca announced a $2 billion equity investment in Summit on September 28, representing approximately 12% of outstanding common stock, alongside collaboration studying Summit’s ivonescimab with AstraZeneca antibody-drug conjugates. Each retains rights to its own medicines. AstraZeneca, September 28

Inference: Financing and combination trials create development options, not proof of clinical success.

Uncertainty: Closing and conversion conditions, trial efficacy and safety, regulatory approvals, and commercial returns remain uncertain.

Why it matters: Track dilution, cash runway, trial milestones, cost sharing, and commercial rights rather than extrapolating from a share-price reaction.

Quant & Market Dashboard

Liquidity and funding

  • Futures: Dow −0.07%, S&P 500 −0.04%, Nasdaq-100 +0.06% at 4:06 a.m. CT. This indicative snapshot precedes the ETF indications below by over two hours. Consolidated spreads, depth, and volume were unavailable.
  • ETFs: SPY was indicated at 767.00, +0.18%, at 6:33 a.m. CT; QQQ at 738.80, +0.31%, at 6:24 a.m. CT. These are non-consolidated indications, not executable NBBO quotes. September 28 closes were 765.61 and 736.53, on 40.78 million and 40.97 million shares respectively. Volume is activity, not a depth estimate. SPY, QQQ
  • Treasuries: September 28 indicative par yields: two-year 4.92%, ten-year 5.24%, twenty-year 5.60%, thirty-year 5.56%; daily increases of 11, 7, 6, and 7 basis points. These are prior-session observations, not live order-book data. U.S. Treasury
  • Funding: Latest verified observations were September 25 SOFR at 3.90% and EFFR at 3.88%. September 28 fixings were not verified at the cutoff. SOFR, EFFR
  • Volatility: September 28 VIX closed at 16.07. October, November, and December VX settled at 17.7083, 18.4798, and 18.8250—contango. No live spot-VIX reading is asserted. Cboe VIX, VX settlements

Volatility regime

Moderate, medium confidence; elevated opening and quarter-end execution risk. Mid-teens VIX and VX contango are non-stressed. Monday’s yield jump, broad factor weakness, oil pressure, and AI-policy headlines nevertheless warrant fresh implied/realized-volatility and liquidity checks.

Factor proxies

Factor ETF September 28 Four weeks*
Momentum MTUM −0.91% +5.07%
Value VLUE −1.45% −1.40%
Quality QUAL −0.32% +0.11%
Size IWM −0.69% −4.73%
Growth IWF −0.83% +2.03%
Low volatility SPLV −0.25% −4.78%

September 28 versus September 25 and August 31 closes. Split-adjusted ETF prices excluding dividends; investable proxies, not pure-factor total returns. All six declined Monday: low volatility and quality resisted best, value lagged. Momentum led over four weeks; low volatility and size were weakest.

Execution implications

  • Separate premarket, opening auction, first 30 minutes, policy headlines, and quarter-end closing flows; do not presume a known imbalance direction.
  • Require current spreads, depth, quote age, and auction imbalances. Missing liquidity cannot justify assumed fills.
  • Use auction-aware limits, participation caps, cancellation controls, and realized-slippage kill switches.
  • Refresh SPY/QQQ volatility, skew, open interest, and strike liquidity. Dealer gamma direction is unverified; 0DTE effects depend on actual positioning.
  • Reject stale, crossed, zero-bid, one-sided, or thin option quotes. Version venue-specific quote controls ahead of October 1 and verify the affected-symbol list.

Watchlist: SPY, QQQ, IWM, TLT, VIX, MTUM, VLUE, QUAL, IWF, SPLV, NVDA, META, AZN, SMMT, XLE, USO, ES, NQ, CL, ZT, ZN, ZB, VX.

DRL Improvements for the MATLAB System

  1. Add session, quarter-end, policy-headline, NQ–ES strength, curve-change, VX-slope, quote-age, and venue-rule-version states.
  2. Enforce drawdown, concentration, turnover, liquidity, and daily-loss constraints outside the reward; penalize downside risk, non-executable fills, and implementation shortfall within it.
  3. Randomize joint equity/bond shocks, depth loss, spread widening, rejected quotes, partial fills, latency, fees, and nonlinear impact.
  4. Use purged, embargoed walk-forward evaluation with frozen baselines, multiple seeds, regime scorecards, and untouched event and rule-change blocks.

Model hygiene

Control leakage, survivorship and look-ahead bias, publication timestamps, corporate actions, stale options quotes, crossed markets, and multiple testing. Preserve first-known data and rule vintages; calibrate out of sample, retain rejected trials, and monitor feature, probability, policy, cost, and regime drift.

MATLAB optimization

  • Align market, macro, auction, and venue-rule data with timetable, retime, and synchronize; retain first-publication timestamps.
  • Vectorize rolling features, preallocate episode buffers, and profile environment steps before parallelizing.
  • Run parallel bayesopt across independent purged folds with reproducible random streams and multiple policy seeds.
  • Reserve GPUs for sufficiently large neural-network batches; separate environment, execution-cost, constraint, venue-rule, and agent modules.

High-value research question

Does a technology-led premarket rebound after a joint equity/bond selloff persist through QQQ’s first hour when conditioned on NQ–ES strength, the two-year-yield change, and VX slope, after auction and 0DTE costs?

Test against an opening-gap baseline. This separates cross-asset information from calendar effects and execution friction.

Research and education only—not individualized financial advice.

If this helped your research, forward it to one colleague who works on markets, models, or execution.

Read more